A computed tomography-based clinical-radiomics model for prediction of lymph node metastasis in esophageal carcinoma
Li Xu, Qingwei Liu, Beini Hu, Jingxu Xu, Chencui Huang, Fang Liu
Abstract
Aims: Evaluation of lymph node metastasis (LNM) is an essential component of preoperative assessment of esophageal carcinoma (EC). This study aimed to develop and validate a computed tomography (CT)-based clinical-radiomics model for the prediction of LNM in patients with EC. Subjects and Methods: This is a retrospective study of 195 patients with biopsy-proven EC. 70% of the included patients were randomly allocated to the training cohort and the remaining 30% of subjects were allocated to the testing cohort. Radiomics models were developed based on features of multi-phase contrast-enhanced CT images using the least absolute shrinkage and selection operator method. The predictive values of these models for LNM were examined in both the training and testing cohorts. Furthermore, the benefits of adding two clinical features (CT report of LNM and tumor location) to the models were also investigated. Results: Seven radiomics models were established based on features identified on single-phase images (plain, P; arterial phase, A; and venous phase, V) and multi-phase images (P + A, P + V, A + V, P + A + V). The model that included 26 features derived from P + A + V had the best predictive value in the training cohort (area under the receiver operator characteristic curve [AUC] 0.783) and testing cohort (AUC: 0.741). The inclusion of CT reports of LNM to the models further improved their performances (AUC 0.814 in the training cohort and AUC 0.813 in the testing cohort). Conclusions: A clinical-radiomics model based on a multi-phase CT study may be useful in predicting LNM in EC.